Turnitin AI Report Metadata: What It Shows in an Appeal
The AI Writing Report shows the detector result. Submission Details identify the submission. Mixing the two makes an appeal harder to check. This guide maps each field to the correct Turnitin surface and explains what the evidence can and cannot prove.
HumanPen Team
· 6 min read
What is actually in the AI Writing Report?
If you are looking for the submission ID, submission date, file name or word count in a Turnitin AI Writing Report, the first thing to know is that Turnitin does not document those as AI-report fields. It documents them under Submission Details in the Similarity Report. The AI Writing Report has a narrower job: it shows the AI indicator, an overall percentage, a submission breakdown, highlighted passages and report states.
That distinction matters in an appeal. You may need both surfaces, but they answer different questions.
Turnitin's current AI Writing Report guide names these parts:
| AI Writing Report item | What it records | What it does not establish |
|---|---|---|
| AI writing indicator | whether a result is available and the displayed state | who wrote the paper |
| Overall percentage detected as AI | the share of qualifying prose classified as likely AI-generated or AI-modified | the share of every word in the file |
| Submission Breakdown | how the detected category contributes to the overall result | a verified history of which tool was used |
| Highlights and interactive bar | where the report places detected passages in the submitted document | a word-by-word probability or exact passage boundary |
| Indicator state | 0%, *%, 20-100%, loading, processing error, ineligible file, or detection disabled at submission | a misconduct decision |
Turnitin also says the AI percentage is independent of the Similarity Score. The AI highlights are not shown in the Similarity Report. The two reports may sit in the same viewer, but they are not one dataset.
Where are the submission ID, date and word count?
Turnitin lists them in Submission Details in the new Similarity Report. An instructor opens that panel with the information icon in the report's upper-right corner. The exact information available can depend on the institution's license.
| Similarity Report surface | Officially documented fields |
|---|---|
| Submission tab | Class ID, Class Name, Assignment, Submission ID, Submission Date, Submission Count, Student ID |
| File tab | Author Name, Last Modified by, Paper Size, Software, File Name, File Extension, File Size, Character Count, Word Count, Page Count |
| Student Similarity PDF | Submission date, submission ID and word count, plus the selected current-view layers |
This is more than a naming detail. Calling all of those values "AI-report metadata" makes it easy to attach a field to the wrong artifact. In an appeal letter, say where each value came from: "Submission ID shown in Similarity Report Submission Details," not "the AI detector says my submission ID is..."
Older exported report PDFs may repeat shared submission details in page headers or footers. That still does not turn those details into AI-classification fields. Label the artifact and the surface where the value is officially documented.
Can a student see the same AI report as an instructor?
No. Turnitin's AI detection FAQ says the indicator and report are visible only to instructors and administrators, not students. It also says an instructor can download the AI report as a PDF and share it with a student.
If you have only been told a percentage, ask for the complete AI report PDF rather than a cropped screenshot. A complete copy preserves the indicator state, the breakdown and the highlighted passages together. Separately request the relevant Submission Details or Similarity Report download if you need the submission ID, date or word count.
Turnitin documents the student Similarity PDF as a snapshot of the current view. Active layers affect what it contains. That is another reason to label each file clearly and avoid treating one export as a complete copy of every Turnitin surface.
How to assemble a report-based appeal evidence packet
An appeal packet is strongest when each artifact has one defined job.
| Artifact | Question it can help answer | Important limit |
|---|---|---|
| Complete AI Writing Report PDF | What did the detector display, and which passages did it highlight? | It does not decide authorship or misconduct. |
| Submission Details or Similarity PDF | Which submission, date and file context are under review? | File metadata can be ambiguous and license-dependent. |
| Original submitted file | Is this the same artifact that was assessed? | The file alone does not show how it was drafted. |
| Version history, drafts and notes | How did the work develop over time? | A history may be incomplete or lack access permissions. |
| Assignment instructions and AI policy | What assistance was allowed for this task? | A detector cannot interpret the course policy for you. |
| Sources and citation records | Can the claims and quotations be traced? | Source use does not by itself settle authorship. |
Use the Submission ID and submission date to anchor the packet to the correct attempt. Keep the original filename and file copy. Then place the AI report beside the process evidence rather than asking the AI score to carry the whole case.
Two metadata fields that are easy to overread
The Similarity Report may display Author Name and Last Modified by for supported files. Turnitin warns that Author Name is the file creator name and can belong to a parent, peer or institution. "Last Modified by" is the last person to open and change a DOCX before submission. Neither label is a reliable answer to "Who wrote this paper?"
A mismatch may have an ordinary cause: a university Microsoft Office license, a shared computer, a parent who checked spelling or a final edit on another device. Record the mismatch and explain it with supporting evidence. Do not call it proof of tampering, and do not assume a matching name proves authorship either.
The same caution applies to word count. The Word Count in Submission Details describes the file. The AI percentage uses qualifying prose as its denominator. Those are different quantities, so multiplying the file word count by the AI percentage does not produce a reliable count of flagged words.
Do not mix in Authorship or Clarity records
Turnitin describes Authorship as a different technology that uses metadata and forensic language analysis to look for possible non-student authorship. It does not identify whether text was AI-written.
Turnitin Clarity is different again. Its Writing Report can show a writing-process overview, paste activity, playback and timeline data, and AI chat interactions when enabled. Students see a Clarity Writing Report only if the instructor allows it for that assignment.
Those records may matter in a case, but they are not fields inside the AI Writing Report. Name the product surface every time you cite one.
A short request you can send
Please share the complete AI Writing Report PDF for the submission under review, along with the Submission Details or Similarity Report page showing the submission ID and submission date. I would like to make sure my response refers to the correct submission and the complete report rather than a cropped score.
This request does not argue that the detector is right or wrong. It asks for the minimum record needed to understand what was reviewed.
The practical rule
The AI Writing Report tells you what the detector displayed. Similarity Report Submission Details identify the submission and file context. Version history, drafts, notes and sources describe the writing process. Keep those layers separate, then connect them with the submission ID and date.
That gives an instructor or academic-integrity reviewer a record they can check. A percentage on its own cannot do that.
KEEP READING
Turnitin AI Writing Report vs Similarity Report: do not read the scores as the same thing
11 min read
Turnitin reportsFalsely flagged for AI: how to appeal, and what evidence your university will accept
22 min read
Tools & workflowHow to keep version history in Word, Google Docs and Overleaf
19 min read